Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 16 papers Another 265 papers do not have category tags yet and are not included in filtered results.
“
About this collection Collection in progress We are building a large AI safety paper collection to help researchers find work related to their interests. Papers are collected automatically andfiltered only for relevance to AI safety , without judging quality, novelty or reading value. The default “Annotated first” shows papers with tags or published paper analyses first. Each group is ordered by original publication date, newest first, with undated papers last. “Newest first” uses publication dates only. This order is not a quality ranking.
One-line summaries, tags and paper analyses are prepared by AI from the papers. Language models may have systematic biases and produce hallucinations.We plan to carry out human annotation and review to reduce errors in this content.
提出 LADE，用首 token 暗知识在生成前过滤越狱查询
Published Oct 6
Evaluated models Llama-2-7B-Chat , Llama-3-8B-Instruct , Qwen2-7B-Instruct and others13 Abstract 作者用 LADE 在生成前过滤有害查询，并称八个模型中六个的 Held-out 平均最好。LADE 是一种只读首 token 输出概率的防御，用来在生成前挡下可能引出有害回复的查询。
提出收缩约束 SSM 安全头，认证嵌入扰动并检测越狱
Published Oct 2
Evaluated models ToySSMClassifier , S4 safety head Abstract 收缩给出玩具 LTI 的有界认证。给 SSM 安全头一个收缩条件，并在越狱检测上使用收缩正则的 S4 头。 。
提出 MOMAT 多图集检索与轻量判定，防御量化模型越狱
Published Oct 1
Evaluated models Llama2-13B , Llama2-7B , Mistral-7B 提出推理时 Unicode 归一化的检测器，识别遭字符级扰动的 AI 生成文本
Published Oct 1
Evaluated models microsoft/deberta-v3-large , deberta-v3-base , bert-large-cased and others4 Abstract 推理时归一化消除两类 Unicode 攻击，训练时归一化则会删掉对抗监督。DeBERTa-ConPara 是一个面向部署的 AI 文本检测器：DeBERTa-v3-large 在多数据集原始文本上训练，推理前做确定性 Unicode 归一化，用一个固定阈值面对没有目标域标签的场景。
构建 TriggerBench 并训练 ThinkingGuard，检测图文隐式风险
Published Sep 29
Evaluated models ThinkingGuard（Qwen3-VL-8B-Instruct） Abstract TriggerBench 隔离触发条件，ThinkingGuard 用分步归因在隐式风险检测上取得作者报告的较高平均分。这篇工作研究多模态隐式风险检测 ：图文各自良性，风险只在关键元素与触发元素交互时出现。
提出 HE-Guardrail，在密文推理中评估越狱护栏并门控回复
Published Sep 18
Evaluated models Llama-3-8B-Instruct , Llama Guard 3-8B Abstract 密文护栏门控 HE-LLM 回复。把越狱护栏装进同态加密 LLM 推理：服务器不解密，也能按安全决策决定客户端看到目标回复还是拒绝回复。
提出 RemTraceNet，少样本识别图像是否经过指定去除流水线
Published Sep 18
Evaluated models RemTraceNet 用后验重加权解释多模态上下文越狱，并提出风险门控的良性示例注入防御
Published Sep 8
Evaluated models Qwen3-VL-2B , Qwen3-VL-4B , Qwen3-VL-8B and others8 Abstract 后验重加权解释上下文越狱的缩放，并支持风险门控的良性示例防御。Zhang 等把多模态 ICL 越狱写成对齐模型在安全与有害隐模式之间的推理时后验漂移，并据此做风险门控防御。
提出 SRD-GUARD，以语义改写与多模型联合评分拦截包装式越狱
Published Sep 6
Evaluated models Llama-3-8B-Uncensored , DeepSeek-V4-Flash 评估安全监视器并提出双侧梯级微调以减少会答有害请求的漏检
Published Sep 5
Evaluated models Gemma-4-31B-IT , Qwen3-32B , Llama Guard 3 8B and others6 Abstract 论文发现安全监视器大多拦截模型已拒答请求，指出明确度偏差是共性原因，并通过双侧梯级微调提升回答请求的召回表现。
提出生成式护栏 HiveTraceGuard-Pro，检测提示注入、越狱与混淆
Published Sep 1
Evaluated models HiveTraceGuard-Pro (0.6B) Abstract 0.6B 俄英护栏在自建俄语干净集和 PI-RU 上得分最高，聚合 key 居第三，过阻断和回复精度仍是作者标出的短板。HiveTraceGuard-Pro 是面向请求与回复审核的 0.6B 生成式护栏，由 Qwen3-0.6B 经 LoRA 微调，在俄语和英语上训练，只输出一个 safe/unsafe 裁决。
提出 FoodGuardBench 与护栏 FoodGuard-4B，评测并降低食品危害越狱
Published Apr 2
Evaluated models Claude-3.7-Sonnet , GPT-4o , GPT-4.1 and others10 提出代理层 peak + accumulation 公式，检测多轮分发的攻击
Published Feb 12 微调编码器抽取语言特征并分类，检测临床训练对话中的越狱
Published Feb 11
Evaluated models BERT-Base , BERT-Large , DistilBERT and others7 提出 HASTE，挖掘 hard negative 再训练恶意提示检测器
Published Jan 27
Evaluated models protectai/deberta-v3-base-prompt-injection Abstract HASTE 用 hard negative 再训练检测器。HASTE 是检测器侧的闭环硬化框架：持续构造能绕过当前检测的对抗提示，再用来更新恶意提示分类器。
提出多轮多模态越狱，并用 FragGuard 拦截有害响应
Published Jan 9
Evaluated models LLaVa-1.6 (7B) , LLaVa-1.6 (13B) , Qwen-2-7B and others5 You have reached the end